The Genome Recombination/Regulation Section focuses on two topics related to recombination and genome stability: mechanisms that generate DNA palindromic gene amplifications and the origin of DNA synthesis errors associated with genetic recombination. Studies on genome instability. We are continuing our analysis of how DNA palindromes are generated. These head to head DNA sequences are highly unstable. Some tumor cells undergo gene amplification by unknown mechanisms that generate palindromes. The instability of these sequences contributes to additional genome rearrangements that occur in tumors. Because palindromes are unstable in bacteria, it is it nearly impossible to clone them. Similarly, the secondary structures that can be adopted by palindromic DNAs make them very difficult to sequence. We opened the field of research on the origin of DNA palindromes by making progress in three important areas related to the study of palindromes. First, we identified yeast strains that tolerate palindromes. Second, we developed a method that allows us to sequence palindromic DNAs. Third, we developed a recombination substrate that generates palindromes and identified a class of recombinants that is almost exclusively palindromes. We demonstrated that the palindromes are formed in our system by a novel kind of nonhomologous end joining (NHEJ) which is independent of some of the recombination functions that are required for most NHEJ events. We recently demonstrated that we can isolate palindromic sequences from mammalian genomes, opening the door to the analysis of palindromes found in normal and malignant cells. We are collaborating on the analysis of DNA palindromes and inverted repeats found in human tumors. This common mechanism of gene amplification in tumors was not accessible to physical characterization until the breakthroughs we made described above. We have developed new methods to isolate the novel junctions associated with DNA palindromes found in tumors. It is our expectation that the characterization of those junctions will help reveal details of the mechanism by which they are generated. Our similar approach to DNA palindromes in yeast was paradigm shifting in that it revealed a very different mechanism of formation quite unlike the generally accepted model. This year we applied high throughput sequencing to the fast annealing fraction of a human tumor cell line. This so called Cot0 DNA includes highly reiterated DNAs and inverted repeats. Our focus is on the inverted repeat fraction. Our analysis demonstrates that some regions of the genome are present as inverted repeats in the human breast tumor cell line MCF7 which are not inverted repeats in normal human cells. This analysis paves the way for us to isolate the novel junctions found in these inverted repeats to test the hypothesis that these structures result from a foldback priming mechanism. Our demonstration that foldback priming causes similar inverted repeats in yeast was the first new discovery of this pathway. Our goal is to investigate whether this mechanism represents an important pathway for gene amplification in tumor cells. Many of the new generation of techniques used to sequence DNAs involve PCR steps. Our research this year demonstrated that this approach systematically excludes the recovery of DNA palindromes because of the requirement for strand displacement when amplifying sequences that can form secondary structures. We generated a unique substrate with a 2 kb DNA palindrome and showed that current second generation sequencing approaches fail on that substrate. These results clearly show that the current in silico representation of the human genome was generated with techniques that could fail to accurately or completely determine the content and structure of the human genome. We are now developing new approaches to solving this problem that will allow the identification and sequencing of DNA palindromes in normal and malignant cells. We initiated a collaborative project with PacBio and the ATP Sequencing Facility to evaluate that approach to sequencing DNA palindromes. These experiments had two important results. First to demonstrate that the software used by PacBio could not readily recognize the novel junctions in palindromes. Second, when that problem was fixed, it is clear that PacBio can successfully sequence DNA palindromes. We are engaged with PacBio in extending these results to provide a general method of identifying and sequencing DNA palindromes in human tumor samples.